424806 - WARNING: leaking reference to nsTimerImpl

Status: NEW

(Core :: XPCOM, defect)

Description

[Bob Clary [:bc] (inactive)]

WARNING: leaking reference to nsTimerImpl: file c:/work/mozilla/builds/1.9.0/mozilla/xpcom/threads/nsTimerImpl.cpp, line 464

Found with the JavaScript tests:

<http://test.bclary.com/tests/mozilla.org/js/js-test-driver-standards.html?test=js1_5/extensions/regress-336409-1.js;language=type;text/javascript> : linux|mac|win32

<http://test.bclary.com/tests/mozilla.org/js/js-test-driver-standards.html?test=js1_5/extensions/regress-336410-1.js;language=type;text/javascript> linux|mac|win32

<http://test.bclary.com/tests/mozilla.org/js/js-test-driver-standards.html?test=js1_5/Regress/regress-3649-n.js;language=type;text/javascript> win32

Comment 1

[Bob Clary [:bc] (inactive)]

These tests also give:

WARNING: An event was posted to a thread that will never run it (rejected): file c:/work/mozilla/builds/1.9.0/mozilla/xpcom/threads/nsThread.cpp

Does that need a new bug?

Comment 2

[Jeff Walden [:Waldo]]

The first warning is bogus given that if nsTimerImpl::PostTimerEvent fails, the weak pointer to the timer that's stored in the created event is released (from the fix for bug 420521).  The warning is only valid if the timer runs, but it's not here; we should remove it.

The second is the real problem, but it's not obvious from the code what's wrong -- real question is who's posting the event to a dead thread, since that's the only way to trigger the second warning.  It's possible the OOMing nature of those tests plays a factor here, but it's hard to tell exactly how.